本文展示了一种自动识别视频中移动目标的方法。论文中提取移动目标通过帧序列，这种方法不需要先验知识，比如：时间阈值调整。基于相邻帧的连续对称差分，我们能得到全分辨率显著图；然后利用最大熵方法计算阈值决定候选区域和获得兴趣点的种子；最后用修改的模糊生长方法获得最终的结果。本文中提出的算法是有效的、具有鲁棒性的。实验结果也证明它具有很好的效果。(This paper presents a method for automatic recognition of moving targets in video. In this paper, moving target is extracted through frame sequence. This method does not require prior knowledge, for example, time threshold adjustment. Continuous symmetry of adjacent frames based on the difference, we can get the full resolution salient map; then calculate the threshold decision candidate region and obtain seed points of interest using the maximum entropy method; finally, using modified fuzzy method to get the final result of growth. The algorithm presented in this paper is effective and robust. Experimental results also show that it has very good results)